Spatial Analyses Methodology

What These Analyses Do

Spatial analyses identify geographic clusters of high and low values using Local Indicators of Spatial Association (LISA). This approach reveals whether geographic proximity correlates with outcome similarity and identifies hotspots, coldspots, and spatial outliers across Australian statistical areas.

Approach

The analysis uses Moran's I statistic to quantify spatial autocorrelation; whether the value of a variable in one location is correlated with values in nearby locations. Spatial weights are defined using Queen contiguity based on ABS statistical area boundaries, meaning locations sharing an edge or corner are considered neighbours.

Statistical significance is assessed using permutation-based inference, testing whether observed spatial clustering is unlikely under the null hypothesis of random spatial distribution.

LISA Classification

Areas are classified into four categories. Hotspots have high values surrounded by other high-value neighbours. Coldspots have low values surrounded by low-value neighbours. Spatial outliers are areas whose values differ substantially from their neighbours; either high values in low-value surroundings or vice versa.

Important Caveat

Results are sensitive to the geographic scale of analysis; a phenomenon known as the Modifiable Areal Unit Problem (MAUP). Spatial clustering detected at one scale may vanish or reverse at different scales. Interpret spatial patterns as specific to the chosen geographic level.

Data Source

ABS Census 2021, AUSynth v1.0.